Qualitative Election Study of Britain Party Leader Evaluations Database, 2005-2024

Winters, Kristi and Carvalho, Edzia and Oliver, Thom (2025). Qualitative Election Study of Britain Party Leader Evaluations Database, 2005-2024. [Data Collection]. Colchester, Essex: UK Data Service. 10.5255/UKDA-SN-857826

This is the latest version of this item.

The Qualitative Election Study of Britain (QESB) Party Leader Evaluation Database 2010 – 2024 contains the collection of 5,504 words and phrases that evaluate British political party leaders collected during pre-election focus groups and interviews conducted with participants from England, Scotland and Wales, during the General Election campaigns of 2010, 2015, 2017, 2019, and 2024.

The aim of this component of the QESB was to get language in use data on the various party leaders, to suppliment, compliment and perhaps contradict closed-ended question results from public polling and social surveys.

These data aren't released alongside the transcript data, however the transcripts refer to respondents' answers. The PIs of the study decided to collect the nearly 15 years worth of raw data and transform it for secondary reuse.

Data description (abstract)

In each round of the Qualitative Election study, participants were asked to provide the words and phrases they associate with British political party leaders. The data were collected during pre-election focus groups and interviews conducted with participants from England, Scotland and Wales, during the General Election campaigns of 2010, 2015, 2017, and 2019. For the in person data collection, party leader evaluation pre-election component, participants were provided with head shot pictures of the party leaders (depending on where in Britain they lived) taken from the party’s own website. In 2024 the data collection was moved online and participants filled in the words and phrases as part of the application process to join the study.

Participants were instructed to write down words or phrases they associate with each person, and indicate if that association was positive, negative or neutral. During the focus groups, the focus group moderator would lead the group in a discussion of the positives, negatives and neutral qualities of each leader. The data included in this dataset only includes the words and phrases written down by the participants, and not the subsequent discussion.

These data are structure for use in sentiment analysis. Each tab contains a column listing participant’s’ words and phrases as a string variable; the next two columns list the election year and leader, affective evaluations (relating to, arising from, or influencing feelings or emotions) as a string variable, and the affective evaluation as a numeric scale from negative –1 to positive +1. These data are suitable for sentiment and discourse analysis, or analytic generalization – establishing that a concept exists within a population regardless of the number of people who hold it.

The new version contains the 2005 and 2014 supplement data.

Words from the 2005 British General Election are substantively different from the other data. They were collected from the article 'Hearts or Minds: Men, Women and Leader Evaluations in the 2005 General Election' by K Winters, R Campbell in Political Communications: The General Election Campaign of 2005, 184-202. Words coded as 1 indicate the respondent thought the word was important in their considerations.

In addition, a supplementary dataset is provided of leaders’ evaluation data from a study with residents of Dundee after the Scottish Independence Referendum in 2014 (N = 287).

Data creators:

Creator Name	Affiliation	ORCID (as URL)
Winters Kristi	GESIS Leibniz Institute for Social Sciences	http://0000-0001-9405-8059
Carvalho Edzia	University of Dundee	http://0000-0001-9441-5829
Oliver Thom	University of the West of England	https://orcid.org/0000-0002-6879-0550.

Contributors:

Name	Affiliation	ORCID (as URL)
Cheron Lou	University of Dundee
Meier Max	University of Würzburg
Gold Cerys	University of Dundee
Challis Phoebe	University of the West of England, Bristol

Sponsors:

The British Academy, The Carnegie Foundation, UK Data Archive, GESIS Leibniz Institute for the Social Sciences, University of Dundee, UWE Bristol

Topic classification:

Politics

Keywords:

POLITICAL LEADERS, POLITICAL ATTITUDES, PARLIAMENTARY ELECTIONS, BRITISH POLITICAL PARTIES, ELECTIONS, REFERENDUMS, ELECTION DATA

Project title:

Cumulative Qualitative Election Study of Britain 2010–2024 Party Leader Evaluations Database, Plus 2005 and 2014 Supplement

Grant holders:

Dr. Kristi Winters, Dr. Edzia Carvalho, Dr. Thom Oliver

Date published:

27 May 2025 09:07

Last modified:

27 May 2025 09:07

Coverage and Methodology

Collection period:

Date from:	Date to:
2005	2024

Geographical area:

England, Scotland, and Wales

Country:

United Kingdom

Spatial unit:

No Spatial Unit

Data collection method:

The data in this dataset were sourced from the Qualitative Election Study of Britain. Focus groups and interviews were conducted in England, Scotland and Wales pre- and post-election. The 2010 –2015 participants were recruited into the QESB panel using a convenience sample from e-mail solicitation plus referrals, and from 2017, Facebook ads were used for recruitment. The PIs over-recruited in each wave and invited participants according to a quota to achieve diversity in partisan affiliation, age, geographic location and employment. To diversify the participant pool, top-up interviews were used to collect data from voters who could not attend an in-person focus groups. Further, online focus groups were used for the same reason. Online participants completed the evaluation exercise before their focus group or interview using a Word or Google document. The leaders evaluation pre-election exercise was conducted in the same way in each wave, except for 2024 as already noted. Participants were provided a sheet with head shot pictures of the party leaders (depending on where in Britain they lived). Party leader pictures were taken from the party’s own website. Participants were instructed to write down words or phrases they associate with each person, and indicate if that association was positive, negative or neutral.

Observation unit:

Text unit

Kind of data:

Numeric, Text

Type of data:

Qualitative and mixed methods data, Time series data

Resource language:

English

Access and Administration

Data sourcing, processing and preparation:

The data structure mirrors data structures used in sentiment analysis. Each tab contains a column listing participant’s’ words and phrases as a string variable;, the next two columns list the election year and leader, affective evaluations (relating to, arising from, or influencing feelings or emotions) as a string variable, and the affective evaluation as a numeric scale from negative –1 to positive +1, except for 2005 where 1 indicates the content is important in the participant's view of the leader. The data processing team made notes on unusual or double coded words or phrases in a separate tab called Outliers. Decisions on how outliers were coded are documented there. n.b. QESB participants were recruited for theoretical reasons – i.e. voters’ attitudes and opinions – and not for representativeness, therefore statistical generalization is not possible from these data. Further, some participants contributed several words, while others contributed fewer, therefore simplistic frequency distribution comparisons cannot be made.

Rights owners:

Name	Affiliation	ORCID (as URL)
Winters Kristi	GESIS Leibniz Institute for the Social Sciences	http://0000-0001-9405-8059
Carvalho Edzia	University of Dundee	http://0000-0001-9441-5829

Contact:

Name	Email	Affiliation	ORCID (as URL)
Winters, Kristi	k.winters@gesis.org	GESIS Leibniz Institute for the Social Sciences	http://0000-0001-9405-8059
Carvalho, Edzia	edziacarvalho@gmail.com	University of Dundee	http://0000-0001-9441-5829